refs https://github.com/TryGhost/Team/issues/1724
- wires trial days stored on a tier to stripe checkout session creation
- removes deprecated `trial_from_plan` if trial days is set
refs https://github.com/TryGhost/Team/issues/1716
- Adds the bio field to the API output
- Allow setting bio when updating the member
- Includes new E2E tests for the members API that were missing
refs https://github.com/TryGhost/Team/issues/1709
- New event type `comment_event` (comments and replies of a member in the activity feed)
- Includes member, post and parent relation by default
- Added new output mapper for ActivityFeed events
**Changes to `Comment` model:**
* **Only limit comment fetched to root comments when not authenticated as a user:**
`enforcedFilters` is applied to all queries, which is a problem because for the activity feed we also need to fetch comments which have a parent_id that is not null (`Member x replied to a comment`). The current filter in the model is specifically for the members API, not the admin API (so checking the user should fix that, not sure if that is a good pattern but couldn’t find a better alternative).
* **Only set default relations for comments when withRelated is empty or not set:**
`defaultRelations`: Right now, for every fetch it would force all these relations. But we don’t need all those relations for the activity feed; So I updated the pattern to only set the default relations when it is empty (which we also do on a couple of other places and seems like a good pattern). I also updated the comments-ui frontend to not send ?include
refs https://github.com/TryGhost/Team/issues/1717
- Updates last_commented_at and last_seen_at (only once a day)
- Used the LastSeenAtUpdater, so we can combine updating last_commented_at and last_seen_at in one query + used same pattern
- Updated comments service to await emails in order to make E2E tests more stable (as we don't have any method to await emails and test emails otherwise). This removed the email sending logic from the `onCreated` hook of the model.
refs https://github.com/TryGhost/Team/issues/1674
- While preparing the changes had a look around and made small refactors to understand the codebase a little better. In general it's best to keep the method parameters as small and precise as possible instead of passing around a "bag-of-all-the-things" like "data" around
refs https://github.com/TryGhost/Team/issues/1674
- While preparing the changes had a look around and made small refactors to understand the codebase a little better. In general it's best to keep the method parameters as small and precise as possible instead of passing around a "bag-of-all-the-things" like "data" around
refs https://ghost.slack.com/archives/C02G9E68C/p1652980792270029
When bulk unsubscribing members, the number of deleted newsletter relations are returned instead of the number of members with newsletters that were cleared. This update deletes newsletter relations on member_id, so we can return the count of members instead of newsletter relations that were deleted.
Tests in https://github.com/TryGhost/Ghost/pull/14871
no issue
- Setting a dedicated source for imported members was already done with 455778662c
- This unifies the same source but keeps the usage of context setting on init alive
refs https://github.com/TryGhost/Team/issues/1577
The call to `edit` was not loading the newsletter relations which is needed
by the serializer used by the webhooks service.
Co-authored-by: Fabien "egg" O'Carroll <fabien@allou.is>
no issue
Using a designated source for members being added within the `importer` context will better describe the source in our members event table. It's more consistent with the usage of the other possible sources.
no issue
When importing members, the members-importer isn't aware of the context and therefore falls back to use `member` as a source in members event table. This makes it impossible to determine imported members from others.
- Added an `context` property to the options in the members-importer constructor
- Checked for `importer` context when creating member events and assigned the source `admin` to it
refs https://github.com/TryGhost/Team/issues/1478
- When creating or editing a member, we'll add the newsletter ids to the subscribe events
- When susbcribing to multiple newsletters, we'll create multiple subscribe events
- When removing newsletters from a member, we'll create one or more subscribe events
- Tests in https://github.com/TryGhost/Ghost/pull/14579
refs https://github.com/TryGhost/Team/issues/1519
- Removed offer metadata again from offer (added in one of the previous commits)
- Added `getByStripeCouponId` method in offer repository (required to find an offer based on the stripe_coupon_id)
- Match discounts from Stripe based on the stripe_coupon_id instead of metadata
refs https://github.com/TryGhost/Team/issues/1519
- Added offer repository dependency to member repository (offerAPI didn't work because it creates a new transaction that resulted in a deadlock during tests)
- Store the offer id from the Stripe subscription metadata in the subscription (only if the discount is still active)
- Also added the offer id to the metadata for a Stripe coupon, this will make adding and removing coupons a bit more foolproof
- Prefer the usage of the offer metadata from a coupon if it is present
- When no discount is applied to a subscription, it always sets the offer id to null, even when the metadata still contains the offer
- The offer_id remains stored when a subscription is canceled/expired
refs https://github.com/TryGhost/Team/issues/1469
Previously, members were subscribed to all available newsletters by default when added. This change updates the default newsletters subscription for member to take into account newsletter preferences for auto opt-in as well as visibility.
refs: https://github.com/TryGhost/Toolbox/issues/293
Things needed to create this:
* MemberSubscriptionEvent now has an import source
* Importer now creates events with this type
* Verification trigger logic changed to use 30 day window of imports
* Updated member mrr_delta calculation to use stored mrr
refs https://github.com/TryGhost/Team/issues/1456
Simplifies the calculation of MRR deltas, which will make it easier to update MRR to include offers and cancellations in the future.
* Adjusted MRR and MRR delta calculation to consider "forever" duration offers
refs https://github.com/TryGhost/Team/issues/1451
Uses the discount information from Stripe to calculate the MRR (this was the easiest way to include it + also supports manually created discounts from users)
refs https://github.com/TryGhost/Team/issues/1456
- Sets the `mrr` column correctly when updating a subscription
- We can use this value in the future to have an improved calculation of mrr_delta, but we first need to have the migration in place
- Updated missing cancelled -> canceled
refs https://github.com/TryGhost/Team/issues/1469
When a new member signs up on the site, by default they should get subscribed to all newsletters. A new member can get added via different ways (stripe webhook, admin add, free signup), so this change updates the base member repository which gets used irrespective of how member is added to ensure default newsletters are always included for new member.
If a member already has a custom newsletters list attached, we don't change anything.
refs https://github.com/TryGhost/Team/issues/1387
First, this default value didn't make sense as it was written when we
had a boolean visible column, second, we don't need to add defaults here
- this implementation is super tied to the models anyway, so we may as
well rely on the model defaults - if we do pull it out similar to
Offers, then it may make sense to add a default there.
We also update the JSDoc to match the updated data structure.
refs: 23b383bedf
- @tryghost/error constructors take an object, not a string - the expectation is that message, context & help should all be set
- This does the bare minimum and just ensures message is set correctly
refs https://github.com/TryGhost/Team/issues/1319
Rather than allowing our code to attempt to speak with Stripe and error
there, we opt to fail fast and throw validation errors so that our API
will respond with a 422, the existing pattern for validation errors.
no-issue
We were incorrectly mixing transactional and non-transactional
operations. An e2e test in Ghost will be merged shortly which caught
this problem.
refs https://github.com/TryGhost/Team/issues/1299
- Convert `created_at` to the right column in each function query
- Renamed the misspelled `getEmailDeliveredEvents` function
- Updated existing unit tests to cover the order
refs: https://github.com/TryGhost/Toolbox/issues/166
New package handles the email verification workflow to prevent spammers. It currently handles MembersSubscribeEvent to detect potential abuse of the API to add members, and exposes methods for checking the threshold / starting the verification process for use by other areas of the code (at the moment - just member imports).
The import package no longer needs to handle anything related to verification since it can be handled in the wrapper function in Ghost, and the API package doesn't need to do anything other than dispatch the new event.
refs https://github.com/TryGhost/Team/issues/1277
- When a user signs-up, two events are created, the code was only keeping one of these events.
- This was introduce in commit 58c9c1c649 when the only usage of the function was to extract the 5 most recent events. Any duplication was creating too much noise.
- This was creating issues now that we introduce event filtering. Some `newsletter_event` events would appear from nowhere we we were filtering-out `signup_event` events.
- We removed the deduplication when the `membersActivityFeed` flag is enabled.
refs https://github.com/TryGhost/Team/issues/1277
- In `getEventTimeline` we filter to only perform the relevant queries, passing to each query function the filters (subset of NQL)
- In each query function, we rewrite the filters to adapt them to the internal data shape.
- We need to do this rewrite to allow API consumers to create filters based on the output on the API instead of the internal data structure.
- Added partial unit tests as there is a lot of repetition between the query functions.
refs https://github.com/TryGhost/Team/issues/1277
- This will allow to filter events within `getEventTimeline`
- The subset of NQL has the following rules:
- Only one level of filters, now parenthesis allowed
- Only three filter keys allowed
- No `or` allowed outside of the bracket notation (this is allowed: `type:-[email_opened_event,email_failed_event]` but this isn't: `type:1,data.created_at:1`)
- The return is an object with a NQL filter by allowed filter key
refs https://github.com/TryGhost/Team/issues/1252
Although we filter out archived tiers from being shown in Portal - we
must also persist this information, so that when they are unarchived,
they continue to not be shown in Portal.
Unfortunately this information is stored in a setting, rather than on
the Tier object itself, two things to consider for the future are:
1. Representing the display value on the Tier object in the API
2. Updating the DB tables to allow storing the display value on the Tier
refs https://github.com/TryGhost/Team/issues/1277
- Doing a filtered DB query wasn't a good solution because we query several tables with different shapes.
- Filtering in memory is done with the NQL library
- Removed the side-effect of changing options.filter for each query, now that options.filter will already contain NQL
- The filtering is done in the reduce function to avoid looping one more time over the in-memory events
refs https://github.com/TryGhost/Team/issues/1277
- Added the pre-existing framework filters to each event query.
- Made sure we aren't modifying the original `options.filter` to avoid side-effects between event queries.
refs https://github.com/TryGhost/Team/issues/1252
This will be used to archive and unarchive Tiers. There is a restriction
on archiving "free" Tiers because our current system expects only one,
and it should always be active.
refs https://github.com/TryGhost/Team/issues/1277
- The members weren't included in the serialized version of the new events
- Some properties weren't using the bookshelf `get` method as they should have
refs https://github.com/TryGhost/Team/issues/1277
- Adds 3 new requests to the `email_recipients` table in the `getEventTimeline` method
- This allows to extract new member events from the table: `email_delivered_event`, `email_opened_event`, `email_failed_event`
refs https://github.com/TryGhost/Team/issues/1037
Tiers have a new `type` column to differentiate between `free` and `paid` tiers. This change -
- sets type as paid for all new tiers created, as `free` tier is created by default
- excludes any price/stripe data change for free tier
- updates all usages of default product to fetch the first paid product from the products list in DB instead of just the first product it finds.
refs https://github.com/TryGhost/Team/issues/1275
We want to be able to track where member subscriptions came from, so
that we can use the information to reduce spam imports of members.
We were missing information when members were uploaded via the Admin
API, and setting the source to 'member' be default - this fixes that
both when creating members and when updating their subscription status.
refs https://github.com/TryGhost/Team/issues/1092
refs https://github.com/TryGhost/Team/issues/1135
This was missed in the initial due to the issue tracking the task being
superceded, and the task not being copied across to the superceding
issue.
A new method to remove coupons has been added, as opposed to updating
the existing change subscription price method, because the removal of a
coupon is not the concern of an auxillary stripe service, but a busines
concern that should be explicit in the members-api codebase.
no-issue
Without forcing linkSubscription to run inside a transaction - it's
possible to have race conditions where it is called twice, and attempt
to insert duplicate rows into the database.
refs https://github.com/TryGhost/Team/issues/1132
We have to include the Offer on the metadata for the Stripe Checkout -
as Offers with a duration of 'once' will not always be present on the
Subscription after fetching it.
Once we receive the Stripe Checkout webhook we emit an event for
subscription created - the reason we use an event is because this logic
should eventually live in a Payments/Stripe module - and we'd want to
decouple it from the Members module.
The Members module is in charge of writing Offer Redemptions - rather
than the Offers module - because Offer Redemptions are "owned" by a
Member - and merely reference and Offer. Eventually Offer Redemptions
could be replaced by Subscriptions.
refs https://github.com/TryGhost/Team/issues/1156
Because we were only attempting to add the product to the members if the
subscription was new AND active - we would not add it for incomplete
subscriptions transitioning to active.
Instead we always attempt to add the product to a member for an active
subscription - it doesn't matter if it's a new one. We later have logic
to filter out duplicate products if the member already has access to the
product.
refs https://github.com/TryGhost/Team/issues/1057
This method will validate a token, and then return the member associated
with it. Rather than exposing token validation and coupling consumers to
the structure of the token response data.
refs https://github.com/TryGhost/Team/issues/995
Since we reintroduced the comped status, we did not update the
subscription handling to correctly set members to a status of comped
when they were on a 'Complimentary' plan.
no-issue
Since updating the product repository to force transactions, the options
parameter was used in every call, meaning it wasn't optional any more,
which broke usage. This updates the parameter to have a default so that
existing usage still works.
no-issue
Since we run our product repository methods in transactions now we must
ensure that all database interations in the method use the transaction.
This adds the missing options to the reading of existing prices so that
they happen inside of the transaction.
refs https://github.com/TryGhost/Team/issues/982
These calls to the edit method were missing the transaction option from
the parent which meant that they ran outside of the transaction and
would cause the method to timeout.
refs https://github.com/TryGhost/Team/issues/979
This correctly handles updates to subscriptions so that if the product a
subscription is for has changed, we will remove the previous product, if
and only if there is not another subscription which gives access to it.
refs https://github.com/TryGhost/Team/issues/873
This handles the creation of product events when a members access to
products is changed. This can happen on creation, update, and any
changes to stripe subscriptions.
We manually workout the difference between the current products and the
new products, and add the events accordingly.
no-issue
Calling ObjectId doesn't return a string a but an ObjectId object.
Whilst this object is cast to a string via the toJSON and toString
methods, this is not enough for MySQL. Instead we should explicitly cast
this to a string ourselves and the application level.
refs https://github.com/TryGhost/Team/issues/946
This adds the bulk edit method which handles bulk edit operations to members
to be used by the filtering feature. They have been combined into a single method
as that is how they are exposed to the API. This is definitely a candidate for a
refactor in the form of a service in front of the repository.
refs https://github.com/TryGhost/Ghost/commit/1dd52075
- Fixes bulkDestroy being passed the context
- Fixes passing options.search to the model layer
- Updates return value since the changes in referenced commit
no-issue
The logic for bulk destroy is currently incorrectly inside of the
members api controller in Ghost core. Moving it out to here allows us to
simplify the controller to rely on the service, rather than implement
the logic.
refs https://github.com/TryGhost/Team/issues/959
Because we were using the pre-existing products to determine a members
status, instead of the products _after_ we have handled the updates to
subscriptions, members with a paid subscription which was later canceled
were changed to 'comped' rather than 'free'. This adds a final check to
set a member to 'free' if their new set of products is empty.
refs https://github.com/TryGhost/Team/issues/919
As we pass the `benefits` to the Product model on creation, we do not
need to manually fetch them again. In fact doing so causes a strange SQL
error, where we attempt to run `SELECT undefined.*`.
refs https://github.com/TryGhost/Team/issues/542
Importing members with a created_at date will incorrectly create events
for the member for the date of the import. This updates our event
handling to use either the passed created_at date, or in the case of
subscriptions the start_date of the subscription. We're using start_date
for subscriptions rather than created, as this is more accurate because
start_date works correctly for backdated subscriptions in Stripe.
refs https://github.com/TryGhost/Team/issues/790
We were missing a check for the existence of memberData.products before
attempting to read the length property from it, which can result in an
Uncaught TypeError
refs https://github.com/TryGhost/Team/issues/790
When linking a subscription to a member we must also update their
status. We could be in one of many states, so we start with the initial
values of either 'free' or 'comped' based on whether the member has any
products. We then make sure to updated the status to 'paid' if we find
any active subscriptions associated with the member, otherwise we leave
it as the initial value.
refs https://github.com/TryGhost/Team/issues/790
Both creating and updating members only ever need to explicitly set
either the 'comped' or 'free' status as these methods do not deal with
Stripe. When updating a member if the products are not changing, we do
not attempt to change the status either.
refs https://github.com/TryGhost/Team/issues/748
This ensures that you cannot add more than one product to a Member.
However it does allow a Member which already exists with more than one
Product to continue using the API. This is to account for edgecases such
as a Member going through the Stripe flow twice and ending up with
multiple subscriptions for multiple products
